DMRS: an efficient dynamic multi-keyword ranked search over encrypted cloud data

With the rise of cloud computing, data owners outsource their data to public cloud servers while allowing users to search the data, aiming at greater flexibility and economic savings. For privacy considerations, private data must be encrypted before outsourcing, and this makes the method of plaintext keyword search infeasible. However, it is critical to enable encrypted data able to be searched. Considering the requirements of practical application scenarios, the function of efficient multi-keyword ranked search and similarity search based on relevance score is necessary for data users. There proposed a number of multi-keyword searchable encryption schemes to try to meet this demand. However, most existing schemes do not satisfy required dynamic update simultaneously. In this paper, a novel and efficient dynamic multi-keyword ranked search scheme improved from traditional secure kNN computation is proposed. The proposed scheme incorporates the similarity measure “coordinate matching” and “inner product similarity” to improve the relevance of search keywords to the relevant cloud files. A reverse data structure is introduced to allow users to perform dynamic operations on document collection, either inserting or deleting. The sparse matrix is used to replace the dense large-scale matrix in index encryption and query vector encryption to improve efficiency. Experiments show that the proposed scheme indeed reduces the overhead of computation and storage compared to MRSE scheme, concurrently guaranteeing privacy and efficiency.